Instructions to use jasperai/flash-sd3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use jasperai/flash-sd3 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-3-medium", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("jasperai/flash-sd3") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
wow impressive super fast!!! insanely BAD HANDS
#1
by froilo - opened
froilo changed discussion title from wow impressive super fast!!! to wow impressive super fast!!! insanely BAD HANDS
Sd3 struggles with hands, anatomy and basically anything with humans so it is expected. Also, most likely, the dataset they trained on for the distillation wasn't probably super optimized for sd3, so results are not too good. Things will become better soon, sd3 just came out so yeah and it is a lot more different then sdxl.
The hands are pure SD3 sucking at anatomy










